There's (exactly) seven ways to optimize latency in an LLM applicationplatform.openai.com3 pointsibigio2 years ago